Real-time multi-objective hand posture/gesture recognition by using distance classifiers and finite state machine for virtual mouse operations

Cameras that are connected to computers record sequence of digital images of human hand in order to interpret human posture/gesture. Human hand posture/gesture recognition has been utilized for providing virtual reality mechanism and it is still an ongoing research in human-computer interaction (HCI) community. Virtual reality can be operated on a particular program but it will be more effective if the entire system can be controlled for the sake of generality. Another direction is the applicability of virtual reality in real time. In this paper, we have developed a virtual mouse system that can recognize the pre-defined mouse movements in real time regardless of the context. Our real time hand recognition system is three fold. 1) skin detection, 2) feature extraction and 3) recognition. For recognition, various features with their own objectives are constructed from hand postures and compared according to the similarity measures and the best-matched posture is used as a mouse action to control the cursor of the computer.